Dynamic backward reasoning systems

摘要

In this paper basic, theoretical backgrounds for dynamic backward reasoning systems are presented. Dynamic reasoning is understood here as one which allows for dynamic changes of the knowledge base, i.e., new facts are added to it, while certain no longer true ones are deleted from the current knowledge base. It is argued that many problems can be easier solved by reasoning backwards than by reasoning forwards. A classification of problems into constructive and destructive ones, with regard to the potential efficiency of systematic backward and forward search is introduced. A formal method for describing physical states and sets of states as well as transformation rules (operators) is given briefly; metarules for performing forward- and backward-chained reasoning are also provided. Moreover, the potential equivalence of reasoning forwards and backwards is proved and the advantages of reasoning backwards are pointed out. This paper, primarily aimed at developing a deeper theoretical understanding of reasoning backwards, can provide basic concepts to be incorporated into a specialized planning or expert system for support of knowledge processing in realistic domains.